Haze Thickness Map Generation for Satellite Image Dehazing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing technologies fail to effectively reduce haze in images, particularly in satellite images, leading to degraded visual quality due to atmospheric particles and noise, which affects various applications like climate research and weather forecasting.
Innovation Solution
A computer-implemented method using machine-learning models and a guided filter to generate a haze thickness map, refine it, and apply it to the input image, along with color correction operations to produce a dehazed and color-corrected image, leveraging dark channel information and multi-scale feature extraction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If machine-learning models are used to reduce haze in images, then haze reduction accuracy is improved, but computational resources and processing time increase
Solution Approach 1:
The image processing is divided into multiple stages: dark channel extraction, haze thickness map generation, guided filtering, and color correction. Each stage processes specific features independently, allowing parallel computation and reducing overall computational burden while maintaining high accuracy through specialized processing at each step.
Solution Approach 2:
The dark channel information is extracted and processed beforehand to generate the initial haze thickness map before applying guided filtering. This preliminary processing of critical haze-related features allows the subsequent filtering stage to work with pre-processed data, reducing the computational load during the main dehazing operation.
2Measurement precision
If detailed feature extraction is performed on all image points, then haze thickness estimation accuracy is improved, but processing time increases
Solution Approach 1:
The algorithm processes different regions of the image with appropriate detail levels. The dark channel extraction and feature window operations focus computational effort on areas with significant haze variations, while guided filtering refines results locally based on surrounding context, ensuring accurate haze thickness estimation without uniformly processing every pixel at maximum detail.
Solution Approach 2:
The method extracts dark channel information and processes feature windows for haze thickness estimation, which are the most critical features for dehazing. Rather than performing exhaustive analysis of all image features, the algorithm focuses on the partial set of features (dark channel, haze thickness map) that provide the most significant contribution to haze reduction accuracy.
Data Source
AI summary
Methods, systems, devices, and tangible non-transitory computer readable media for haze reduction are provided. The disclosed technology can include generating feature vectors based on an input image including points. The feature vectors can correspond to feature windows associated with features of different portions of the points. Based on the feature vectors and a machine-learned model, a haze thickness map can be generated. The haze thickness map can be associated with an estimate of haze thickness at each of the points. Further, the machine-learned model can estimate haze thickness associated with the features. A refined haze thickness map can be generated based on the haze thickness map and a guided filter. A dehazed image can be generated based on application of the refined haze thickness map to the input image. Furthermore, a color corrected dehazed image can be generated based on performance of color correction operations on the dehazed image.